Open Journal of Mo dern Hydrology, 2011, 1, 11-22
doi:10.4236/ojmh.2011.12002 Published Online October 2011 (http://www.SciRP.org/journal/ojmh)
Introducing the Mixed Distribution in
Fitting Rainfall Data
Jamaludin Suhaila, Kong Ching-Yee, Yusof Fadhilah, Foo Hui-Mean
Department of Mat hematics, Facul t y of Sci ence, University Teknologi Malaysia, Johor Bahru, Malaysia
Received July 26, 2011; revised August 29, 2011; accepted September 27, 2011
Abstract
Several types of mixed distribution are proposed and tested in order to determine the best model in
describing daily rainfall amount in Peninsular Malaysia for the time period of 33 years. A mixed distribution
is a mixture of discrete and continuous daily rainfall which included the dry days. The mixed distributions
tested in this study were exponential distribution, gamma distribution, weibull distribution and lognormal
distribution. The model will be selected based on the Akaike Information Criterion (AIC). In general, the
mixed lognormal distribution has been selected as the best model for most of the rain gauge stations in
Peninsular Malaysia. However, these results are greatly influenced by the topographical, geographical and
climatic changes of the rain gauge stations.
Keywords: Mixed distribution, Akaike Information Criterion (AIC), Maximum Likelihood Estimator (MLE),
Mixed Lognormal
1. Introduction
Flood and drought are the calamity that can cause by
imbalance amount of rainfall and amount of runoff in the
certain area. Flood happen when the amount of rainfall is
greater than the outflow of water, while drought is vice
versa. Both these disasters will give great impact to ag-
riculture sector and cause death if the disaster is seriously
befallen. Hence, by studying the characteristic of the
rainfall, preparations to overcome these disasters can be
done earlier in order to reduce any lose that occur. Hence,
continuous researches interested on the distribution of
hydrology have been carried out [1-4].
Modelling rainfall data can be distinguished into two
parts: rainfall occurrence and rainfall amount. A model
of rainfall occurrence is a model that provides a sequence
of dry and wet days, while a model of rainfall amounts
simulates the amount of rainfall occurring on each wet
days [5]. Markov chain models are often used to fit the
rainfall occurrence [6,7]. On the other hand, two pa-
rameters gamma distribution, exponential distribution,
weibull and lognormal are among the theoretical distri-
bution used to fit the rainfall attribute [8-13]. However,
gamma distribution only models the amount of rainfall in
wet days [14].
In Peninsular Malaysia, studies on finding the best
model for rainfall data had been carried out by several
researchers. Mixed-exponential is the best fit distribution
for hourly rainfall data among exponential, gamma and
weibull [15]. While mixed lognormal is the most appro-
priate distribution for describing the daily rainfall amount
compare to lognormal and skew normal [16]. Based on
these studies, the mixed distribution is seemed more
suitable in describing rainfall data in Peninsular Malaysia.
Hence, mixed distributions are more suitable for Penin-
sular Malaysia [17,18].
However, past studies that have been conducted in
Malaysia, only considered rainfall amount on wet days,
which does not following the nature of rainfall where
there are days that do not rain at all. The importance of
included the zeros (dry days) is the characterization of
daily rain rate, drought, or climate change effects can by
analyze [19]. To the best of author knowledge, the mix-
tures of two distributions which include the rainfall data
for dry days and wet days have not yet been done in Ma-
laysia. Hence the study proposes to investigate the con-
cept of mixture of these rainfall data.
2. Study Area
Malaysia is located in Southeast Asia which has two land
mass which separated by South China Sea. The land

12 J. SUHAILA ET AL.
mass that on the Asian mainland is called Peninsular
Malaysia and the other is East Malaysia with two states
Sarawak and Sabah which located on the island of Bor-
neo. The area of Malaysia is approximately 330,000
square kilometres and share border with Thailand (in the
north), Singapore, Indonesia (in the south), Brunei and
the Philippines (in the east). The weather in Malaysia is
generally hot and humid due to its location which near to
the equator. The climate between the east and west coasts
are different due to two monsoon seasons that annually
strike in Malaysia. The southwest monsoon occurs from
May to August while the occurrence of northeast mon-
soon is during November to February. The periods be-
tween these two monsoons are named as inter monsoon
seasons. Northeast monsoon usually bring heavy rain to
east coast of Peninsular Malaysia. Compared to the
northeast monsoon, the southwest monsoon is much drier
throughout Peninsular Malaysia due to the Peninsular
Malaysia is protected by Sumatran (Indonesia) mountain
range. During the inter monsoons seasons, the west coast
of Peninsular Malaysia will reach the maximum monthly
mean rainfall. Generally, the annual rainfall in Malaysia
is between the ranges of 2000 to 4000 mm with uniform
temperature which ranged from 25.5˚C to 32˚C through-
out the country.
3. Rainfall Data
The daily rainfall data used in this study were obtained
from the Malaysian Meteorological Department and
Drainage and Irrigation Department which contain the
period of 33 years (1975-2007). Seventy rain gauge sta-
tions were chosen for this study. The quality of rainfall
data was checked through the homogeneity test, which
are the standard normal homogeneity test, Buishand
range test, the Pettitt test and the Von Neumann ratio test
[20]. The stations chosen were scatter around in the area
of Peninsular Malaysia. The details about the stations are
shown in Table 1 and Figure 1.
4. Methodology
4.1. Modeling Rainfall Amount
Most of the data are either discrete or continuous. The
characteristic of rainfall data is neither continuous com-
ponent nor discrete component, but it is a mixture of both
components. However, the rainfall data were often as-
sumed as continuous values in which zero rainfall val-
ues were ignored. A mixed distribution was suggested by
combining the discrete and continuous components [21].
For mixed distribution, given a random sample 1,
X
2,, n
X
X that containing mn

zeroes (dry days), the
likelihood of the random sample with parameter;
and
p

is as follows:


1
,| 1,,
nm
mm
x ppfxfx

Lp



   (1)
where is the total of wet days, and is a parametric
family distribution. Equation (1) does not represent the
true likelihood if the data are dependent. The MLE of p
is given as
mf
ˆ pmn

.
In this study, four mixed distribution model were used
to determine the appropriate model for rainfall charac-
teristic in Peninsular Malaysia. The probability density
function and the logarithm of the likelihood function of
the four distributions will be described with X as the
random sample for each distribution.
 Exponential distribution is given as

e, 0
;0, 0
xx
fx x






 (2)
where 0

 is named as rate parameter or the scale
parameter which determined the variation of rainfall
amount series. By using (1), the likelihood of exponent-
tial distribution is shown below:



1
,1 i
nmm e
x
m
i
Lp xpp




  (3)
Then, solve the log likelihood function and the MLE for
ˆ

is given asˆ1
x
 . The same method is applied for
others distribution in order to find their log likelihood
function and MLE.
 Use e The weibull distribution with two parameters is
described as follows:



1e,
;,
0, 0
x
xx0
x


 




 fx (4)



where 0

 is the shape parameter and 0

 is the
scale parameter. Shape parameter represents the shape of
the distribution and scale parameter determines the
variation of rainfall amount series.
The logarithm of the likelihood function of mixed
weibull is given as below





1
1
ln,,lnln 1
lnln1 ln
i
m
i
i
m
i
i
xmpn mp
mm x
x






 
 



Lp
(5)
To solve the nonlinear equation, the method known as
Simple Iterative Procedure is employed [22].
 The probability density function for gamma distribu-
tion can be written as




1
;, e
x
fx x


 
 


0x
for (6)

J. SUHAILA ET AL.
13
Table 1. The l atit ude and l ongi tud e of t he c hosen 70 st atio ns.
Code Station Name Latitude Longitude
E01 Kota Bharu 6˚10'12"N 102˚16'48"E
E02 To’ Uban 5˚58'12"N 102˚08'24"E
E03 Sek. Keb. Kg. Jabi 5˚40'48"N 102˚33'36"E
E04 Kg. Merang, Setiu 5˚31'48"N 102˚57'00"E
E05 Stor JPS Kuala Trengganu 5˚19'12"N 103˚07'48"E
E06 Kg. Menerong 4˚56'24"N 103˚03'36"E
E07 Klinik Bidan, Jambu Bongkok 4˚56'24"N 103˚21'00"E
E08 Sek.Men. Sultan Omar, Dungun 4˚45'36"N 103˚25'12"E
E09 Sek. Keb. Kemasek 4˚25'48"N 103˚27'00"E
E10 JPS Kemaman 4˚13'48"N 103˚25'12"E
E11 Kuantan 3˚46'48"N 103˚13'12"E
E12 Rumah Pam Pahang Tua, Pekan 3˚33'36"N 103˚21'36"E
E13 Endau 2˚35'24"N 103˚40'12"E
E14 Mersing 2˚27'00"N 103˚49'48"E
NW01 Abi Kg. Bahru 6˚30'36"N 100˚10'48"E
NW02 Guar Nangka 6˚28'48"N 100˚16'48"E
NW03 Padang Katong ,Kangar 6˚27'00"N 100˚11'24"E
NW04 Arau 6˚25'48"N 100˚16'12"E
NW05 Kodiang 6˚22'12"N 100˚18'00"E
NW06 Alor Star 6˚12'00"N 100˚24'00"E
NW07 Ampang Pedu 6˚14'24"N 100˚46'12"E
NW08 Pendang 5˚59'24"N 100˚28'48"E
NW09 SIK 5˚48'36"N 100˚43'48"E
NW10 Dispensari Kroh 5˚42'36"N 101˚00'00"E
NW11 Rumah Pam Bumbong Lima 5˚33'36"N 100˚26'24"E
NW12 Bkt Berapit 5˚22'48"N 100˚28'48"E
NW13 Ldg. Batu Kawan 5˚15'36"N 100˚25'48"E
NW14 Klinik Bkt. Bendera 5˚25'12"N 100˚16'12"E
NW15 Kolam Takongan Air Itam 5˚24'00"N 100˚16'12"E
NW16 Pintu A.Bagan, Air Itam 5˚21'00"N 100˚12'00"E
NW17 Rumah Penjaga JPS. Parit Nibong 5˚07'48"N 100˚30'36"E
SW01 Jam. Sg. Simpangn, Jln. Empat 2˚26'24"N 102˚11'24"E
SW02 Malacca 2˚16'12"N 102˚15'00"E
SW03 Pekan Merlimau 2˚09'00"N 102˚25'48"E

14 J. SUHAILA ET AL.
SW04 Ldg. Bkt. Asahan 2˚23'24"N 102˚33'00"E
SW05 Tangkak 2˚15'00"N 102˚34'12"E
SW06 Pintu Kawalan Separap Batu Pahat 1˚55'12"N 102˚52'48"E
SW07 Pintu Kawalan Sembrong 1˚52'48"N 103˚03'00"E
SW08 Sek.Men.Inggeris Batu Pahat 1˚52'12"N 102˚58'48"E
SW09 Ldg. Kian Hoe, Kluang 2˚01'48"N 103˚16'12"E
SW10 Kluang 2˚01'12"N 103˚19'12"E
SW11 Ldg. Benut, Rengam 1˚50'24"N 103˚21'00"E
SW12 Ibu Bekalan Kahang, Kluang 2˚13'48"N 103˚36'00"E
SW13 Sek.Men.Bkt Besar di Kota Tinggi 1˚45'36"N 103˚43'12"E
SW14 Senai 1˚37'48"N 103˚40'12"E
SW15 Ldg. Getah Kukup, Pontian 1˚21'00"N 103˚27'36"E
SW16 Stor JPS Johor Bahru 1˚28'12"N 103˚45'00"E
W01 Stn. Pemereksaan Hutan, Lawin 5˚18'00"N 101˚03'36"E
W02 Selama 5˚08'24"N 100˚42'00"E
W03 Rumah JPS, Alor Pongsu 5˚03'00"N 100˚35'24"E
W04 Pusat Kesihatan Bt.Kurau 4˚58'48"N 100˚48'00"E
W05 Gua Musang 4˚52'48"N 101˚58'12"E
W06 Ipoh 4˚34'12"N 101˚06'00"E
W07 Ldg Boh 4˚27'00"N 101˚25'48"E
W08 S. K. Kg. Aur Gading 4˚21'00"N 101˚55'12"E
W09 Sitiawan 4˚13'12"N 100˚42'00"E
W10 Rumah Kerajaan JPS, Chui Chak 4˚03'00"N 101˚10'12"E
W11 Rumah Pam Paya Kangsar 3˚54'00"N 102˚25'48"E
W12 Ibu Bekalan Sg. Bernam 3˚42'00"N 101˚21'00"E
W13 Kg. Sg. Tua 3˚16'12"N 101˚41'24"E
W14 Gombak 3˚16'12”N 101˚43'48”E
W15 Empangan Genting Kelang 3˚14'24"N 101˚45'00"E
W16 JPS. Wilayah Persekutu 3˚09'36"N 101˚40'48"E
W17 Genting Sempah 3˚22'12"N 101˚46'12"E
W18 Janda Baik 3˚19'48"N 101˚51'36"E
W19 Sg.Lui Halt 3˚04'48"N 102˚22'12"E
W20 Ldg. Sg. Sabaling 2˚51'00"N 102˚29'24"E
W21 Setor JPS Sikamat Seremban 2˚44'24"N 101˚57'36"E
W22 Hospital Port Dickson 2˚31'48"N 101˚48'00"E
W23 Ldg. Sengkang 2˚25'48"N 101˚57'36"E

J. SUHAILA ET AL.
15
WEST
EAST
SOUTHWEST
NORTHWEST
THE LOCATION OF THE RAIN GAUGE STATIONS IN
PENINSULAR MALAYSIA
Figure 1. The location of the 70 rain gauge stations in Pen-
insular Malaysia.
where 0

 is the shape parameter and 0

 is the
scale parameter for gamma distribution.
The logarithm of the likelihood function of mixed
gamma is given as


 


1
1
ln,,lnln 1
1ln ln
ln 1
i
m
i
i
m
i
i
Lpxmpn mp
xm
m

x





 
 



(7)
where the MLE for ˆ

is ˆˆ
x

  while the MLE for
ˆ

is


1ln ln
m
i
ixm




 which
 


  

is the digamma function.
 The probability density function for lognormal distri-
bution is



22
ln 2
;,1 2πex fx x

 




 for (8) 0x
where

and

are the location and scale parameters
respectively.
4.2. Goodness-of-Fit Tests (GOF)
GOF is used to determine the best model among the dis-
tributions tested in rain characteristic. In this study, AIC
was used to select the best model. The model that attains
the lowest AIC will be the best model among the com-
petitive distribution. The result of AIC is directly de-
pendent with the sample size of observation [23]. AIC is
asymptotically effective and unbiased since the test is
based on the maximum likelihood function and if the
sample size is sufficiently larger than 30, the test will
yield fairly accurate result [24]. The sample size of this
study is greater than 30, hence AIC can be applied to
determine the best model. The formula for AIC is given
as
2ln 2
A
ICL k

 (9)
where is the logarithm of the likelihood function
of the propose model and k is the number of parameters.
ln L
5. Result and Discussion
This section is divided into two main sub-sections. The
descriptive statistics for each of the seventy rain gauge
stations will be discussed in the first sub-section fol-
lowed by a discussion on fitting distributions in the sec-
ond sub-sections.
5.1. Descriptive Statistics
The descriptive statistics in terms of mean, standard de-
viation, coefficient of variations (CV), skewness, the
maximum amount of rainfall and number of wet days of
the annual rainfall amount for each seventy rain gauge
stations are summarized in Tabl e 2. Based on the values
of descriptive statistics, the five highest mean rainfall
amounts among the stations are Chui Chak (W10), fol-
lowed by Kg Menerong (E06), Endau (E13), Selama
(W02) and Pusat Kesihatan Bt. Kurau (W04). Due to the
geographical locations of Selama, Chui Chak and Bt.
Kurau stations which full of limestone bedrock, granitic
hills and mine waste deposits (e.g. slime, tailings and
mining ponds) [25].
Lake is an indicator of high level climate and moun-
tain also can affect the climate in the area [26]. These
situations somehow contribute to the increase in total
amount of rainfall in those areas. Lawin (W01), Ldg.
Sg.Sabaling (W20), Raya Kangsar (W11), Guar Nangka
(NW02) and Sitiawan (W09) are among the stations that
received the lowest mean rainfall. Most of these stations
are at the inland areas of Peninsular Malaysia in which
the climate at these areas is less affected by the mon-
soons. The climate for most of the inland stations is rela-
tively dry [27].
In terms of CV, three of the stations (W12, W10 and
SW09 stations) attain the highest CV among the other
stations which in the range of 28% to 45%. These results
the irregularity of the daily rainfall received by the sta-
tions. On the other hand, the alue of skewness is affect v

16 J. SUHAILA ET AL.
Table 2. Statistic of annual rainfall amount for seventy rain gauge stations.
Code Station Mean CV (%) Skewness Maximum amount
of rainfall (mm)
Number of
Wet Days
E01 2514.17 23.74 5.99 591.5 4378
E02 2711.21 26.72 4.95 409 4632
E03 2782.36 17.83 3.93 329.5 4487
E04 2740.71 20.13 3.97 365.5 3884
E05 2574.25 16.63 5.33 520.4 4623
E06 3572.57 22.77 6.67 676 6245
E07 2491.53 13.6 8.49 790 3814
E08 2492.8 18.43 5.02 572 4599
E09 2610.47 15.78 4.21 330 4365
E10 2662.02 13.7 4.88 434 4796
E11 2910.43 18.71 4.99 527.5 4947
E12 2522.35 16.49 5.04 444.8 4447
E13 3192.49 17.28 3.98 353.5 5313
E14 2624.81 23.22 4.91 383.3 4890
NW01 1891.97 16.77 2.63 125.5 4522
NW02 1741.53 15.86 3.15 218.5 4274
NW03 1895.09 17.4 2.6 150.1 4482
NW04 1967.99 20.17 2.51 180 4389
NW05 1902.01 20.08 2.66 163 4531
NW06 1971.59 16.28 2.56 172.1 4493
NW07 2018.02 26.46 2.79 211.5 4665
NW08 2222.72 21.89 3.12 261 4865
NW09 2594.84 23.61 2.71 220.5 5043
NW10 2062.3
18.37 2.6 175 5084
NW11 2139.28 21.67 3.32 275 4339
NW12 2182.73 26.49 3.79 295 4650
NW13 1837.35 14.96 2.67 206 3872
NW14 2764.42 13.96 4.35 484.8 5199
NW15 2258.85 18.72 3.67 308.5 4917
NW16 2506.14 20.88 2.59 245.5 4270
NW17 2187.5 15.13 2.75 230 3584
SW01 2371.15 23.61 2.44 217.5 4990
SW02 1980.71 17.73 2.86 275.2 4428
SW03 1964.88 14.51 3.22 272.5 4168

J. SUHAILA ET AL.
17
SW04 1769.56 13.27 2.3 172 3550
SW05 1895.45 19.86 2.29 152.7 4366
SW06 1910.17 14.28 2.33 137.5 3963
SW07 2146.75 17.92 3.64 320 4811
SW08 2222.21 18.83 2.86 200.5 4623
SW09 1929.68 29.96 2.27 210 3087
SW10 2116.02 22.68 4.82 433.4 4767
SW11 2199.05 19.49 2.87 210 4550
SW12 2760.32 12.71 4.31 372 5395
SW13 2076.83 22.13 3.64 271.5 5012
SW14 2447.37 11.5 4.21 364.4 5369
SW15 2467.35 20.55 3.39 260 4797
SW16 2407.34 15.92 3.3 313.6 5132
W01 1682.42 14.76 2.82 159 4586
W02 3165.87 13.75 2.01 176 5400
W03 2417.88 17.95 2.52 170 5187
W04 3151.77 16.42 2.21 193 6435
W05 2331.09 22.48 2.57 154.5 5544
W06 2487.51 14.74 2.08 135.4 5248
W07 2187.79 12.55 2.27 110 6481
W08 2315.17 15.07 2.1 130 4077
W09 1760.24 18.42 2.67 178.7 4251
W10 3590.15 31.11 1.36 160.5 4926
W11 1733.74
10.85 3.58 259.8 4314
W12 2548.82 44.89 2.29 173.6 5274
W13 2386.12 20.17 2.39 173.5 5317
W14 2421.35 15.89 2.13 139 5297
W15 2354.28 14.87 2.22 162.5 5233
W16 2577.83 17.34 2.66 289 5425
W17 2483.4 13.88 16.46 800.5 6018
W18 1863.52 24.28 3.47 210 5044
W19 2182.01 20.9 2.3 215 3111
W20 1722.8 13.58 3.19 226 3936
W21 1974.31 16.71 2.37 144.5 4674
W22 1997.41 18.88 2.77 200 4209
W23 2148.78 14.36 4.8 500 3922

J. SUHAILA ET AL.
18
by maximum amount of rainfall received by the stations.
For example, the Genting Sempah (W17) station has the
both highest maximum amount and skewness value.
While in terms of number of wet days, rain always oc-
curred at these areas. In addition, there is study indicated
that the landslide often occur at area of Genting Sempah
which had caused numbers of deaths and injuries [28].
Hence, some studies had been carried out to predict the
landslide hazard [29] and debris flow [30] in Genting
Sempah.
L = MIXED LOGNORMAL
G = MIXED GAMMA
W = MIXED WEIBULL
THE BEST FIT DISTRIBUTION OF THE RAIN GAUGE
STATIONS IN PENINSULAR MALAYSIA
Due to the effect of northeast monsoon, the stations
(E01, E02, E03, E04, E05, E06, E07, E08, E09, E10,
E12, E13, E14, SW12 and W22 stations) that are at the
east coast of Peninsular Malaysia receive high mean
rainfall amount which are more than 2490 mm. Besides,
the stations (NW14, NW15 and NW16 stations) located
in island take in more rainfall amount than the stations
(NW11, NW12 and NW13 stations) in mainland al-
though the location of these stations are in the same
neighborhood. Not only that, the stations (W12, W13,
W14, W15 and W16 stations) situated in urban area also
receive high rainfall amount. Hence, rainfall amount can
be varying according to the surrounding of the area and
monsoons.
Figure 2. The best fit distribution of the 70 rain gauge sta-
tions in Peninsular Malaysia.
5.2. Fitting Distribution Based on AIC Criterion
The results of AIC values are displayed in Table 3 with
the bolded values indicated the lowest AIC. The best fit
distribution of the rain gauge stations are shown in Fig-
ure 2. The mixed lognormal distribution is dominating
other distributions as the best fitting distribution among
the studied stations. 48 stations attained the lowest AIC
values for mixed lognormal followed by 20 stations cho-
sen for mixed gamma to describe their rainfall data. On
the other hand, only 2 stations chose mixed weibull as
the best fitting distribution. Of all distributions, the
mixed exponential was never selected by any of the sta-
tions.
sure of the stations towards southwest and northeast
monsoons could give impact in the result of the fitting of
the distribution. The geographical sites, climatic changes
and topographical of the stations also will be strongly
affect the result.
6. Conclusions
The urge in finding the most appropriate fitting distribu-
tion for daily rainfall amount in Peninsular Malaysia is
always the main interest in particular studies. The con-
cept of included zero values into rainfall data for attain
the best fit distribution in Peninsular Malaysia still
commence among researchers in Malaysia. Several dis-
tributions have been tested and compare in this study to
find the best fitting distribution. The distributions tested
are mixed exponential, mixed gamma, mixed weibull and
mixed lognormal distributions.
Most of the stations that are located at the east coast
obtain mixed gamma distribution as their best model. It
is possible that the distributions of rainfall data at these
stations are influenced by the northeast monsoon flow.
Meanwhile the Chui Chak station (W10) and Sg. Lui
Halt station (W19) are the only two stations acquired
mixed weibull distribution as the most appropriate fitted
distribution. These stations are located at the foot of the
mountain and recorded relatively high mean rainfall
amount, high in coefficient variation, low in skewness
and low in maximum amount of rainfall. These charac-
teristics of rainfall data may possibly suitable for the
chosen fitted distribution.
The mixed lognormal distribution was the preferred
best fitted distribution for the majority of the stations in
Peninsular Malaysia which determined by Akaike In-
formation Criterion. Mixed gamma distribution was the
second favoured distribution followed by mixed weibull
distribution. A big number of the stations in the east
coast of Peninsular Malaysia were identifying mixed
gamma distribution as the most appropriate distribution.
In general, the distinction of elevation and the expo-

J. SUHAILA ET AL.
19
Table 3. The AIC Values of each of the seventy rain gauge stations.
Code Station Mixed LognormalMixed ExponentialMixed Weibull Mixed Gamma
E01 1780.2 1836.12 1819.75 1776.57
E02 1935.18 1966.82 1966.66 1965.91
E03 1859.61 1875.44 1875.71 1903.17
E04 2026.01 2088.15 2077.12 2052.49
E05 2192.14 2302.2 2245.2 2147.35
E06 3618.95 3719.17 3687.61 3591.95
E07 2180.86 2219.33 2221.1 2238.46
E08 2500.94 2590.35 2561.57 2490.82
E09 2531.7 2627.46 2592.58 2523.7
E10 2920.32 3042.57 2992.85 2889.33
E11 3000.91 3085.3 3058.61 2981.49
E12 2804.99 2915.33 2871.41 2778.78
E13 3187.37 3287.68 3246.43 3140.35
E14 2525.02 2586.52 2567.61 2511.59
NW01 1195.13 1221.16 1221.71 1224.53
NW02 1163.38 1194.62 1188.05 1168.29
NW03 1116.84 1126.51 1127.13 1120.45
NW04 1149.66 1172.35 1168.25 1148.9
NW05 1163.8 1182.11 1178.05 1156.64
NW06 1042.91 1051.43 1052.91 1052.21
NW07 1037.51 1048.67 1047.9 1036.71
NW08 1296.41 1313.28 1312.91 1304.15
NW09 1648.16 1676.94 1677.45 1679.63
NW10 2147.8 2178.05 2178.58 2179.74
NW11 1886.3 1896.38 1898.38 1910.33
NW12 2612.54 2639.22 2639.25 2635.19
NW13 2374.93 2378.72 2380.04 2404.06
NW14 2028.27 2053.87 2052.09 2037.62
NW15 2063.86 2095.01 2093.17 2078.96
NW16 1785.7 1798.41 1800.39 1815.03
NW17 2528.99 2558.74 2559.3 2594.77
SW01 2721.3 2739.43 2741.09 2750.8
SW02 2671.64 2705.31 2705.67 2705.72
SW03 2455.27 2465.63 2467.24 2492.7

20 J. SUHAILA ET AL.
SW04 2473.73 2486.84 2488.7 2497.16
SW05 2441.66 2473.87 2474.51 2476.97
SW06 2921.51 2931.74 2931.54 2970.96
SW07 3207.96 3240.12 3241.3 3249.62
SW08 3623.89 3640.73 3642.52 3624.41
SW09 1936.21 1950.77 1951.47 1949.19
SW10 2713.08 2750.38 2745.72 2720.64
SW11 2782.9 2804.38 2806.36 2829.98
SW12 3096.57 3143.79 3125.71 3053.59
SW13 2727.8 2771.99 2757.7 2697.99
SW14 3098.69 3128.57 3124.44 3092.5
SW15 3407.23 3429.43 3428.52 3480.96
SW16 3181.77 3203.63 3200.96 3174.26
W01 2211.82 2243.58 2244.1 2247.44
W02 3745.16 3770.63 3772.11 3813.83
W03 3319.98 3361.06 3357.32 3335.01
W04 4038.2 4047.06 4048.97 4076.01
W05 2801.42 2832.51 2834.45 2857.5
W06 3637.63 3657.88 3659.5 3669.78
W07 3612.07 3642.85 3644.75 3680.64
W08 2894.37 2920.65 2913.8 2965.79
W09 2617.55 2657.16 2656.14 2649.82
W10 2968.49 2990.87 2960.74 2996.41
W11 2842.54 2871.85 2873.85 2900.59
W12 3673.74 3700.92 3699.17 3683.17
W13 3130.36 3166.75 3165.12 3153.22
W14 2785.77 2817.34 2816.9 2809.3
W15 2886.01 2944.01 2930.3 2878.02
W16 3686.22 3731.25 3720.09 3659.66
W17 3458.86 3485.79 3487.67 3520.3
W18 3022.74 3054.46 3051.82 3105.33
W19 2298.48 2303.94 2279.13 2306.98
W20 2528.13 2557.14 2552.91 2524.27
W21 2660.66 2700.23 2694.04 2662.35
W22 2237.97 2261.93 2263.43 2270.58
W23 2236.53 2253.76 2254.63 2253.97

J. SUHAILA ET AL.
21
These stations are greatly influenced by northeast mon-
soon. Mixed exponential distribution is the only distribu-
tion that has not been selected by any of the stations in
describing the rainfall distribution. In conclusion, the
rain gauge stations in Peninsular Malaysia are greatly
swayed by their topographical, geographical sites and
climatic changes which give great disparity on the rain-
fall distribution.
7. Acknowledgements
Authors are faithfully appreciate the generously of the
staff of Malaysian Meteorological Department and Drai-
nage and Irrigation Department for providing the daily
rainfall data for the usage of this paper. The work is fi-
nanced by Zamalah Scholarship provided by Universiti
Teknologi Malaysia and FRGS vote 4F024 from the
Ministry of Higher Education of Malaysia.
8. References
[1] V. P. Singh, “On Application of the Weibull Distribution
in Hydrology,” Water Resources Management, Vol. 1,
No. 1, 1987, pp. 33-43. doi:10.1007/BF00421796
[2] R. T. Clarke, “Estimating Trends in Data from the Wei-
bull and a Generalized Extreme Value Distribution,” Wa-
ter Resources Researc h, Vol. 38, No. 6, 2002, 1089.
doi:10.1029/2001WR000575
[3] P. K. Bhunya, R. Berndtsson, C. S. P. Ojha and S. K.
Mishra, “Suitability of Gamma, Chi-Square, Weibull, and
Beta distributions as Synthetic Unit Hydrographs,” Jour-
nal of Hydrology, Vol. 334, No. 1-2, 2007, pp. 28-38.
doi:10.1016/j.jhydrol.2006.09.022
[4] P. K. Swamee, “Near Lognormal Distribution,” Journal
of Hydrologic Engineering, Vol. 7, No. 6, 2002, pp.
441-444. doi:10.1061/(ASCE)1084-0699(2002)7:6(441)
[5] T. G. Chapman, “Stochastic Models for Daily Rainfall in
the Western Pacific,” Mathematics and Computers in Si-
mulation, Vol. 43, No. 3-6, 1997, pp. 351-358.
doi:10.1016/S0378-4754(97)00019-0
[6] S. Deni, A. Jemain and K. Ibrahim, “Fitting Optimum
Order of Markov Chain Models for Daily Rainfall Occur-
Rences in Peninsular Malaysia,” Theoretical and Applied
Climatology, Vol. 97, No. 1, 2009, pp. 109-121.
doi:10.1007/s00704-008-0051-3
[7] R. D. Stern and R. Coe, “A Model Fitting Analysis of
Daily Rainfall Data,” Journal of the Royal Statistical So-
ciety. Series A (General), Vol. 147, No. 1, 1984, pp. 1-34.
doi:10.2307/2981736
[8] N. T. Ison, A. M. Feyerherm and L. Dean Bark, “Wet
Period Precipitation and the Gamma Distribution,” Jour-
nal of Applied Meteorology, Vol. 10, No. 4, 1971, pp.
658-665.
doi:10.1175/1520-0450(1971)010<0658:WPPATG>2.0.
CO;2
[9] T. A. Buishand, “Some Remarks on the Use of Daily
Rain-Fall Models,” Journal of Hydrology, Vol. 36, No.
3-4, 1978, pp. 295-308.
doi:10.1016/0022-1694(78)90150-6
[10] W. May, “Variability and Extremes of Daily Rainfall
during the Indian Summer Monsoon in the Period
1901-1989,” Global and Planetary Change, Vol. 44, No.
1-4, 2004, pp. 83-105.
doi:10.1016/j.gloplacha.2004.06.007
[11] H. K. Cho, K. P. Bowman and G. R. North, “A Com-
parison of Gamma and Lognormal Distributions for Cha-
racterizing Satellite Rain Rates from the Tropical Rainfall
Measuring Mission,” Journal of Applied Mete- orology,
Vol. 43, No. 11, 2004, pp. 1586-1597.
doi:10.1175/JAM2165.1
[12] S. R. Bhakar, A. K. Bansal, N. Chhajed and R. C. Purohit,
“Frequency Analysis of Consecutive Days Maximum
Rain-Fall at Banswara, Rajasthan, India,” ARPN Journal
of Engineering and Applied Sciences, Vol. 1, No. 3, 2006,
pp. 64-67.
[13] D. A. Mooley, “Gamma Distribution Probability Model
for Asian Summer Monsoon Monthly Rainfall,” Monthly
Weather Review, Vol. 101, No. 2, 1973, pp. 160-176.
doi:10.1175/1520-0493(1973)101<0160:GDPMFA>2.3.
CO;2
[14] H. Aksoy, “Use of Gamma Distribution in Hydrological
Analysis,” Turkish Journal of Engineering and Environ-
mental Sciences, Vol. 24, No. 6, 2000, pp. 419-428.
[15] Y. Fadhilah, M. Zalina, V. T. V. Nguyen, S. Suhaila and
Y. Zulkifli, “Fitting the Best-Fit Distribution for the
Hourly Rainfall Amount in the Wilayah Persekutuan,”
Jurnal Teknologi, Vol. 46, No. C, 2007, pp. 49-58.
[16] J. Suhaila and A. A. Jemain, “Fitting Daily Rainfall
Amount in Malaysia Using the Normal Transform Dis-
tribution,” Journal of Applied Sciences, Vol. 7, No. 14,
2007, pp. 1800-1886.
[17] J. Suhaila and A. A. Jemain, “Fitting Daily Rainfall
Amount in Peninsular Malaysia Using Several Types of
Exponential Distributions,” Journal of Applied Sciences
Research, Vol. 3, No. 10, 2007, pp. 1027-1036.
[18] J. Suhaila and A. A. Jemain, “Fitting the Statistical Dis-
tribution for Daily Rainfall in Peninsular Malaysia Based
on AIC Criterion,” Journal of Applied Sciences Research,
Vol. 4, No. 12, 2008, pp. 1846-1857.
[19] E. Ha and C. Yoo, “Use of Mixed Bivariate Distributions
for Deriving Inter-Station Correlation Coefficients of
Rain Rate,” Hydrological Processes, Vol. 21, No. 22,
2007, pp. 3078-3086. doi:10.1002/hyp.6526
[20] J. B. Wijngaard, A. M. G. K. Tank and G. P. K. Nnen,
“Homogeneity of 20th Century European Daily Tem-
perature and Precipitation Series,” International Journal
of Climatology, Vol. 23, No. 6, 2003, pp. 679-692.
doi:10.1002/joc.906
[21] B. Kedem, L. S. Chiu and G. R. North, “Estimation of
Mean Rain Rate Application to Satellite Observations,”
Journal of Geophysical Research, Vol. 95, No. D2, 1990,
pp. 1965-1972. doi:10.1029/JD095iD02p01965
[22] H. Qiao and C. P. Tsokos, “Parameter Estimation of the

22 J. SUHAILA ET AL.
Weibull Probability Distribution,” Mathematics and
Computers in Simulation, Vol. 37, No. 1, 1994, pp. 47-55.
doi:10.1016/0378-4754(94)90058-2
[23] R. P. McDonald, “An Index of Goodness-of-Fit Based on
Noncentrality,” Journal of Classification, Vol. 6, No. 1,
1989, pp. 97-103. doi:10.1007/BF01908590
[24] F. M. Mutua, “The Use of the Akaike Information Crite-
rion in the Identification of an Optimum Flood Frequency
Model,” Hydrological Sciences Journal, Vol. 39, No. 3,
1994, pp. 235-244. doi:10.1080/02626669409492740
[25] B. K. Tan, “Urban Geology of Kuala Lumpur and Ipoh,
Malaysia,” International Association of Engineering Ge-
ology (IAEG2006), The Geological Society of London,
London, 2006, pp. 1-7.
[26] C. E. Vincent, T. D. Davies and A. K. C. Beresford,
“Recent Changes in the Level of Lake Naivasha, Kenya,
As an Indicator of Equatorial Westerlies over East Af-
rica,” Climatic Change, Vol. 2, No. 2, 1979, pp. 175-189.
doi:10.1007/BF00133223
[27] W. L. Dale, “The Rainfall of Malaya, Part II,” Journal of
Tropical Geography, Vol. 14, 1960, pp.11-28.
[28] Z. A. Roslan and A. H. Zulkifli, “'ROM Scale for Fore-
casting Erosion Induced Landslide Risk on Hilly Ter-
rain,” In: Kyoji Sassa, Ed., Landslides: Risk Analysis and
Sustainable Disaster Management: Proceedings of the
Fist General Assembly of the International Consortium
on Landslides, Part 3, International Consortium on Land-
slides, Springer, Berlin Heidelberg, 2005, pp. 197-202.
[29] S. Ahmad, M. I. S. Mohd and M. Hashim, “Application
of Remote Sensing Techniques for Prediction of Land-
slide Hazard Areas in Malaysia,” 14th United Na-
tions/International Astronomical Federation Workshop
on “Capacity Building in Space Technology for the Bene-
fit of Developing Countries,” with an Emphasis on Natu-
ral Disaster Management, Vancouver, 2-3 October 2004.
[30] B. K. Tan and W. H. Ting, “Some Case Studies on Debris
Flow in Peninsular Malaysia,” In: H. Liu, A. Deng, and J.
Chu, Eds., Geotechnical Engineering for Disaster Miti-
gation and Rehabilitation, Springer, Berlin, 2008, pp.
231-235. doi:10.1007/978-3-540-79846-0_20